The dawn of service management

by John Morency

December 19, 2000 —

The very nature of enterprise network management is about to undergo a fundamental change. In the future, network managers will be focusing less on the traditional tasks of designing, implementing and operating complex LANs and WANs, and more on the selection and management of external service providers.

This shift from network management to service management will require net managers to deal with a whole new set of strategic decisions, such as what parts of the network should be kept private and what parts should go public, and whether voice or data services are to be included. Network managers will also have to negotiate complex service-level agreements (SLA) and monitor them in an environment in which adequate tools to do so don't exist yet.

In the long run, however, network managers will benefit from this shift toward service providers. There will be an unprecedented level of choice in determining support responsibility for the next generation of network and application services. Some, most or all of the services can continue to be managed by the enterprise support staff or turned over to one or more external providers.

The key factors driving shops in the direction of service providers are the explosion in e-business and the difficulty in attracting and retaining qualified staff. For example, Forrester Research predicts that Internet services revenue from business-to-business communications will grow from nearly $4 billion in 1998 to nearly $60 billion by 2003.

And service providers are responding to this anticipated surge in demand. Forrester Research indicates that while worldwide equipment spending for enterprise networks exceeded that for service providers by a substantial 68% to 29% margin, this gap is expected to be narrowed by a factor of three over the next two to three years.

But simply building the infrastructure and offering the services will not suffice. As recent incidents at Amazon.com, eBay and ESPN's Fantasy Baseball site have demonstrated, network outages -- once accepted as a fact of life -- can cost companies millions of dollars and are no longer acceptable.

In addition, network managers are becoming increasingly sensitive to the lost revenue opportunities that occur when poor Web site performance directly results in substantially reduced access traffic. The increasing sophistication of Web management tools is enabling many of these managers to more reliably correlate the occurrence of increased site response time with substantial drops in the number of concurrent online users.

The result is that the once intracompany enterprise network is rapidly becoming the multicompany interenterprise network when e-business becomes the game. And this change not only affects data, but also increasingly voice and video, as well. Consequently, high-quality service tethering of each to the other will become fundamental to the growth of the "E-conomy."

As this happens, traditional monitoring tools that only work in enterprise networks or service provider networks, but not both, will become increasingly inadequate. New tools that support both arenas will become a mandatory reequirement, given the rapidly blurring boundary between public and private networks.

These products are not available yet. But network managers should expect to see product advances beginning in a few key areas over the coming year, most notably for basic Internet access, virtual private networks, digital subscriber line access and managed application services.

The advances will support much more of a shared service management model between the service provider and the enterprise manager than has traditionally been the case. The enterprise manager will benefit as more service providers open their management reporting systems to customer Web browser access.

Public service-level management

A key part of service management is the SLA, a formal contract between service providers and users on the quality of service (QoS), availability and performance that will be delivered by the provider.

Without question, the service provider community is beginning to understand and embrace this shift toward guaranteed levels of service. Today, AT&T, MCI WorldCom and Sprint all offer a minimum of 99.9% (and in a number of service instances 100%) availability for public frame relay, ATM and IP access services. In addition, ISP Concentric Network now offers a 100% availability guarantee, which includes the network backbone and local loop, as well as an average round-trip latency guarantee of 80 msec and a guarantee that packet loss will not exceed an average of 1% per month.

UUNET offers 100% availability guarantees for its dedicated-line and frame relay Internet access services, which cover UUNET's backbone as well as the customer access circuit. A somewhat lower, but still substantial, service guarantee is offered by PSINet, which claims a 99.5% availability level for its basic Internet access services, including InterFrame and IntraNet.

Even with the increase in service provider SLA guarantees, however, few if any corporate customers have had any means of measuring or validating monthly service quality claims. The obvious major exception is frame relay, for which industry players such as DeskTalk Systems, Concord Communications, Visual Networks, NextPoint Software and Paradyne today offer some form of hardware and/or software product support for more rigorous service-level tracking of carrier-based frame relay services.

These tools address a major loophole in SLAs -- the fact that the service provider measures the level of service, rather than the customer or an independent third party. Traditionally, the frame relay user has been dependent on manual reports produced monthly by the provider. This approach suffers from two drawbacks: The first is lack of reporting timeliness, and the second is the lack of perceived reporting objectivity.

Collectively, the drawbacks have limited the perceived value of provider-defined SLAs, particularly with respect to real-time e-business services.

Enterprise service-level management

The effectiveness of SLAs within the enterprise has also been mixed. Despite the fact that service-level management has been a frequent topic of discussion for the press and industry pundits, business justification for the implementation of service- level management and supporting products has been anything but a straightforward exercise for many users.

Research jointly conducted last year by Renaissance Worldwide and McConnell Consulting showed that the business case for implementing in-house service-level management was weak. Implementation often resulted in increased costs without clearly defined benefits. And the setting of unreasonably high expectations was quickly followed by end-user dissatisfaction with the QoS delivered by IT.

An additional finding was the common realization among many of the users that current product offerings were only focused at lower-level network transport services or hiigher-level network application services, but not at supporting both sets of services within the same product. Therefore, product offerings did not often align well with internal service delivery expectations.

In this context, respondents said they needed service management products that would provide a level of availability and performance support for corporate e-mail systems and Lotus Notes networks comparable to that provided for Fast Ethernet, frame relay or ATM. The benefit, they said, would be a more comprehensive and effective approach to meeting the needs of internal stakeholders without the drawback of excessive product implementation complexity.

Looking at the collective state of existing service provider and enterprise staff-supported SLAs, one finds a complementary set of strengths and weaknesses. In general, service providers have significant technical and staff advantages in service delivery, but oftentimes they have few tools for objective and timely reporting. Numerous product options exist for service-level management within the enterprise, but enterprise support organizations do not often have the technical or support staff to match the level of service delivery offered by an external provider, particularly when Internet and Internet-based applications services are considered.

Come together

What is required, therefore, is a new generation of products that will offer enterprise managers three key features:

A more integrated means of managing the inevitable set of public IP services that will be required to support end-to-end e-business.
The ability to effectively mine enterprise network and provider network-based management data in order to deliver a complete, accurate and timely picture of network performance, delay and availability.
The ability to implement service management from the application to the plumbing layer irrespective of whether application services are being supported in-house or by an external provider.

The good news is that early products sporting some of these features are starting to appear. Leading-edge vendors such as Inverse Network Technologies and Keynote Systems offer network managers the opportunity to benchmark ISP availability and response time for the user's own site, as well as additional sites of the user's choice. In addition, Quallaby Corp. has extended its Proviso product line to enable service providers to deliver Web-based SLA reports to customers.

The products, while certainly breaking some significant new ground, are only the beginning. Over time, users will require increasingly sophisticated service-level management software that supports all network elements involved in the delivery of one or more crucial e-business applications and so blurs the traditional distinction between private and public networks.

In addition, the ability to initiate proactive testing and monitoring of any network component from the enterprise or service provider side will also be important in order to rapidly isolate problems that occur. These are particularly crucial factors for the successful growth of current and future applications service providers.

The stakes are high

The combination of the rapid and profound changes occurring in enterprise network management combined with the mixed success SLAs have had with enterprise and service provider network managers is contributing to the evolution of a new management model -- service management.

Service management refers to the changing role of traditional network managers, who are becoming more like public service portfolio managers, as well as the approach enterprise managers need to take in assembling the components of their

next-generation management tool kits.

Increasingly, the tools will need to concurrently support the management of the in-house network, as well service provider offerrings (particularly those of ISPs) that collectively deliver the set of services for which the enterprise manager is being held accountable. The tools of greatest use will be those that support a shared partnership management model between the service subscriber and the provider in the service-level delivery of one or more e-business applications.

The extent to which this functionality can be accommodated as logical extensions of existing service-level management tools from vendors, such as Concord, DeskTalk, Visual, NextPoint, InfoVista and Micromuse, will enable better progress by more effectively using products that users already have installed.

The difference between this service-level management model and previous ones is the fact that real business is now at stake for both partners. And this alone is incentive enough to ensure that ultimately all the pieces fit together.

Network World